Ashwajit Warwatkar
Built a spiking neural network model of memory replay and consolidation that reproduces key cognitive phenomena including serial position effects. Accepted for oral presentation at BICA 2026 and publication in Springer's Lecture Notes in Electrical Engineering (Scopus indexed). From enrollment to Springer acceptance in 2 months.
Where Ashwajit Started
His Background
- • 12th grader at Alphonsa Sr. Sec. School, Maharashtra
- • Prior research on lung tumor ablation modeling and SCADA systems
- • Explored brain organoid (wetware) computing through independent systematic review
- • Built an affective AI system for astronaut mental health support
- • Experience with Python, ML workflows, and biomedical data analysis
His Goals
- • Publish a peer-reviewed paper in a reputable journal
- • Move from exploratory research to rigorous, publishable work
- • Build a competitive profile for top university admissions
- • Focus on intelligent, adaptive biomedical systems
- • Present at international conferences
The Problem He Wanted to Solve
Memory consolidation, the process by which the brain transforms short-term memories into long-term storage, involves complex interactions between replay, synaptic plasticity, and neural architecture. Existing models treated replay, consolidation, and serial position effects as separate phenomena. Ashwajit wanted to build a unified spiking neural network model that accounts for all three, connecting neuroscience theory with computational simulation.
The Research
Ashwajit developed the Replay-Gated Cascade Consolidation (RGCC) model, a 1,000-neuron Izhikevich spiking network with spike-timing-dependent plasticity and Fusi-type slow cascade weights. He stated three predictions in advance and confirmed all three across 700+ simulation runs, including ablation, causal intervention, parameter scanning, and scale expansion experiments up to 5,000 neurons.
Replay-Gated Cascade Consolidation: A Spiking-Network Account of Replay, Consolidation, and Serial Position Effects
Existing models address replay, consolidation, and serial position effects in isolation, lacking a unified computational framework
1,000-neuron Izhikevich spiking network with STDP, Fusi-type slow cascade weights, coherence-gated inter-memory replay, and schema core architecture
700+ simulation runs across ablation, restoration, causal intervention, null-model, component-ablation, and parameter sensitivity experiments. Scaled to N=2,000 and N=5,000 neurons
Removing replay reduced retention from 0.286 to 0.037 (p<0.001). Harmonic-series prediction fit with R²=0.828. Behavioural read-out showed clear FULL vs. NO_REPLAY separation (0.82 vs. 0.31, p<0.001)
Bridging Neuroscience and Computation as a 17-Year-Old
What makes Ashwajit's work remarkable is its hypothesis-first rigor. He stated three predictions before running any experiments: (P1) inter-memory replay is necessary for retention and depends on the two-timescale architecture; (P2) suppressing replay of early-encoded memories should degrade them while boosting late-encoded replay should not rescue them; (P3) the consolidation gradient should follow a harmonic-series relationship. All three predictions were confirmed across 700+ runs. Reviewers scored the work 5/5 for novelty and significance, and the paper was accepted for oral presentation, the highest presentation tier at the conference.
Presentation Tier
Simulation Runs
Predictions Confirmed
Enrollment to Acceptance
The Outcome
Published in Springer Lecture Notes in Electrical Engineering
BICA 2026 (17th Annual Meeting of the BICA Society)
Springer Lecture Notes in Electrical Engineering
Replay-Gated Cascade Consolidation: A Spiking-Network Account of Replay, Consolidation, and Serial Position Effects
Scopus • EI Compendex
Early-stage research experience with independent projects in biomedical AI and wetware computing, but no peer-reviewed publications
First-author Springer publication on computational neuroscience, accepted for oral presentation at an international conference in 2 months
The Bigger Picture
Years old when he published a spiking neural network model of memory consolidation in Springer
From YRI enrollment to Springer acceptance with oral presentation at BICA 2026
Published in Scopus and EI Compendex indexed proceedings, a credential most undergrads don't achieve
Ready to Start Your Research Journey?
Join the YRI Fellowship and work with expert mentors to conduct original research, get published, and build a profile that stands out.
Apply Now